Optimization of hCFTR Lung Expression in Mice Using DNA Nanoparticles

Abstract

Efficient and prolonged human cystic fibrosis transmembrane conductance regulator (hCFTR) expression is a major goal for cystic fibrosis (CF) lung therapy. A hCFTR expression plasmid was optimized as a payload for compacted DNA nanoparticles formulated with polyethylene glycol (PEG)-substituted 30-mer lysine peptides. A codon-optimized and CpG-reduced hCFTR synthetic gene (CO-CFTR) was placed in a polyubiquitin C expression plasmid. Compared to hCFTR complementary DNA (cDNA), CO-CFTR produced a ninefold increased level of hCFTR protein in transfected HEK293 cells and, when compacted as DNA nanoparticles, produced a similar improvement in lung mRNA expression in Balb/c and fatty acid binding protein promoter (FABP) CF mice, although expression duration was transient. Various vector modifications were tested to extend duration of CO-CFTR expression. A novel prolonged expression (PE) element derived from the bovine growth hormone (BGH) gene 3′ flanking sequence produced prolonged expression of CO-CFTR mRNA at biologically relevant levels. A time course study in the mouse lung revealed that CO-CFTR mRNA did not change significantly, with CO-CFTR/mCFTR geometric mean ratios of 94% on day 2, 71% on day 14, 53% on day 30, and 14% on day 59. Prolonged CO-CFTR expression is dependent on the orientation of the PE element and its transcription, is not specific to the UbC promoter, and is less dependent on other vector backbone elements.

Introduction

Genetic replacement therapy for the lungs of cystic fibrosis (CF) patients has multiple requirements to be effective and safe, including adequate levels of human CF transmembrane conductance regulator (hCFTR) mRNA in proximal lung epithelial cells using a vector system that is nonimmunogenic and nontoxic. Based on in vitro studies, observations in various strains of CF knockout mice, and correlations with human CF lung disease, there is a consensus that achieving ~5–10% of normal levels of hCFTR mRNA may be therapeutic.^1,2,3 Additionally, gene transfer in at least 25% of ciliated proximal lung epithelial cells may be required for airway surface liquid height correction and effective mucus transport.⁴ The therapeutic challenge is to achieve this level of wild-type hCFTR mRNA and efficient gene transfer to the CF lung, where the disease process typically results in thick mucus secretions that may reduce gene transfer efficiencies.

DNA nanoparticles comprised of single molecules of DNA compacted with polyethylene glycol (PEG)-substituted lysine peptides enter the apical membrane of lung epithelial cells by binding to cell surface nucleolin, and this complex then efficiently enters the nucleus of these postmitotic cells.⁵ The small size of the DNA nanoparticles is required for entry through the nuclear membrane pore,⁶ although the size dimension requirements for nuclear uptake in the mouse lung appears somewhat less restrictive than in microinjection studies in cells.^6,7 DNA nanoparticles are nontoxic and do not stimulate CpG responses in the murine lung,⁸ and repetitive dosing to the lungs of Balb/c mice does not reduce transgene expression.⁹ When applied to the nasal mucosa of CF subjects, no adverse events were associated with the nanoparticles and 8/12 subjects had functional evidence of CFTR chloride channel function, with several in the normal range of nasal potential difference testing.¹⁰ The payload plasmid in this initial phase I trial incorporated a cytomegalovirus (CMV) promoter, which rapidly shuts off in the lung,¹¹ and the partial nasal potential difference corrections observed lasted for about 1 week. In total, these encouraging findings suggest that aerosol delivery of compacted DNA nanoparticles may provide sufficient hCFTR expression to address the lung manifestations of CF.

To address anticipated difficulties in transfecting the human CF lung, optimization of the hCFTR expression plasmid to achieve clinically appropriate and prolonged mRNA levels is a premium requirement, since the ability to maintain expression in successively transfected airway cells might address the mobile CF mucus environment. Since airway epithelial cells survive ~200 days before apoptosis,¹² achieving hCFTR transgene mRNA expression for months is the desired objective. To achieve this goal, the DNA nanoparticle plasmid payload must efficiently enter airway cell nuclei and then remain transcriptionally active. Processes that might interfere with long-term hCFTR transgene expression include promoter downregulation,¹³ CpG methylation and/or heterochromatin formation within the transcriptional cassette,¹⁴ and loss of the expression plasmid itself. In this article, we report the successful optimization of hCFTR mRNA expression in the lungs of CF knockout mice by employing multiple strategies to affect the level and duration of hCFTR expression. Levels of hCFTR mRNA and protein expression have been significantly enhanced by the development of a synthetic hCFTR expression “gene ” that optimizes codon utilization, depletes CpG dinucleotides, and improves Kozak consensus sequences. Additionally, prolonged duration of hCFTR expression has been achieved with use of a novel prolonged expression (PE) element derived from the bovine growth hormone (BGH) gene 3′ flanking sequence. In combination, these hCFTR expression plasmids, delivered as compacted DNA nanoparticles to the lungs of CF knockout mice, achieve significant expression levels for multiple months, thereby addressing optimization goals for aerosol lung testing in CF subjects.

Results

Initial vector design and hCFTR gene optimization

We have previously shown that a plasmid (pUL) incorporating the polyubiquitin C (UbC) promoter and a CpG-depleted prokaryotic backbone produces prolonged luciferase activity in the rat and mouse brain,^15,16 and the murine lung (data not shown). The hCFTR gene complementary DNA (cDNA) sequence containing partial 5′ and 3′ untranslated terminal regions (UTRs) was subcloned into the pUL vector (pUCF; Figure 1a). To potentially improve hCFTR expression, a synthetic hCFTR sequence (referred to as CO-CFTR) was designed, synthesized in vitro, and subcloned into the pUL backbone, forming plasmid pUCF2 (Figure 1a). The CO-CFTR is CpG-depleted except for one CpG located close to the 3′ end of the open reading frame; in contrast, hCFTR cDNA in pUCF has 63 CpG dinucleotides. The synthetic CO-CFTR gene has removed the 5′ and 3′ UTRs of natural hCFTR, minimizing potential miRNA affects,¹⁷ and has a 76.9% nucleotide homology to the hCFTR coding sequence. Additionally, CO-CFTR has a 55.6% GC content, whereas hCFTR cDNA has a 41.5% GC content, which may be significant because increases in GC content correlate with transcriptional efficiency.^18,19 To address protein expression efficiency, this synthetic CO-CFTR sequence also was codon optimized, having 85.1% of codons corresponding to the highest codon adaptation index compared to 34.4% of codons with the highest codon adaptation index in natural hCFTR cDNA. Lastly, CO-CFTR has a typical Kozak consensus sequence (CCACCatg), whereas hCFTR cDNA (agACCatg) has only a partial match.

Figure 1.

CO-CFTR generates improved protein and mRNA levels compared to natural hCFTR cDNA. (a) CpG-reduced plasmid vectors incorporating the human polyubiquitin C (UbC) promoter and first intron for human cystic fibrosis transmembrane conductance regulator (hCFTR) expression. pUCF incorporates hCFTR complementary DNA (cDNA) and pUCF2 incorporates CO-CFTR, a codon-optimized, CpG-depleted (except 1 CpG in the 3′ terminus), and 5′ and 3′ untranslated terminal region (UTR) truncated in vitro synthesized DNA.

designates CpG dinucleotides. (b) Immunoprecipitation. (IP)/western blot analysis of hCFTR expression in HEK293 cells 2 days after transfection with pUCF, pUCF2, or control pCMVCFTR plasmid. Cells were transfected with lipofectamine and either low (L, 0.75 µg) or high (H, 3 µg) amounts of CFTR plasmid. Luc, cells transfected with luciferase plasmid; NT, nontransfected. Anti-CFTR monoclonal antibody 1660 (R&D Systems), directed against R domain codons 590–830, was used in both immunoprecipitation and detection protocols. Similar results were noted in three separate experiments. Letter “c ” designates glycosylated forms of hCFTR. (c) hCFTR/mCFTR mRNA levels in Balb/C mouse lungs dosed intranasally (IN) with compacted pUCF and pUCF2 plasmids and harvested at days 2 and 14. (d) hCFTR/mCFTR mRNA levels in fatty acid binding protein promoter (FABP) mice dosed intratracheally (IT) with compacted pUCF and pUCF2 and harvested at day 2. For both (c) and (d), mice received ~100 µg of compacted DNA, solid lines indicate geometric means, and dotted line equals 5%.

CFTR-negative HEK293 cells were transfected using lipofectamine with either pUCF or pUCF2. Two days after transfection, lysates were prepared and an immunoprecipitation/western for CFTR performed (Figure 1b). The western blot showed a ninefold increase in hCFTR protein in cells transfected with pUCF2 compared to pUCF. For reference, cells were also transfected with a CMV promoter-controlled hCFTR cDNA expression plasmid.

hCFTR mRNA expression in the mouse lung: initial steps

Compacted pUCF or pUCF2 was dosed intranasally into Balb/C mice and lungs were harvested at days 2 and 14 for evaluation of CFTR mRNA. hCFTR (or CO-CFTR) mRNA expression is compared to endogenous murine CFTR (mCFTR) mRNA and is presented as the geometric mean of a hCFTR/mCFTR expression ratio multiplied by 100% (see Supplementary Figure S1 and Table S1). As shown in Figure 1c, pUCF2 generated CO-CFTR mRNA expression of 2.34% on day 2 which fell to 0.77% on day 14. Although pUCF2 CO-CFTR mRNA expression was not maintained, CO-CFTR expression levels were significantly higher (14.6-fold) than natural hCFTR expression generated by pUCF nanoparticles (0.16% on day 2) (P = 0.0169, t-test of log-transformed data).

One potential reason for transient hCFTR expression in the Balb/C mouse lung is immunogenicity of hCFTR protein. Screening of SCID and fatty acid binding protein promoter (FABP) (hCFTR expressed in gastrointestinal tract but not lung) mice demonstrated that an immune response to hCFTR is not involved in CO-CFTR silencing (Supplementary Figure S3). Neither of these strains or FABP parental strains (129, C57, and FVB/N) maintained CO-CFTR mRNA expression for two weeks. Since the FABP strain represents a mouse model of CF and produced elevated CO-CFTR expression (167%) on day 2 (Supplementary Figure S3 and Figure 3a) it was chosen as the preferred strain for CO-CFTR expression experiments. All measurements take into consideration the partial knockout of mCFTR in the FABP mouse lung, which generates a 3′ mCFTR partial transcript that is similar in levels to mCFTR expressed in the other tested mouse strains (Supplementary Figure S2 and Table S2). To further compare the efficiency of hCFTR cDNA and CO-CFTR, compacted pUCF and pUCF2 were dosed intratracheally (IT) in FABP mice and harvested on day 2 (Figure 1d). The pUCF2 generated expression was 3.7-fold higher than pUCF (47.9% versus 13.1%; P = 0.0131, t test of log-transformed data).

Figure 3.

CO-CFTR mRNA expression in the FABP lung mediated by compacted pUCF70 is prolonged and related to vector DNA levels. (a) Initial and repeat pUCF70 study (pUCF70′) in fatty acid binding protein promoter (FABP) mice. One hundred and six microgram of compacted pUCF70 were dosed intratracheally (IT) and animals were harvested at days 2, 14, 30, and 59 for quantitative reverse transcription (qRT)-PCR analysis. For comparison are results with pUCF2. Dotted line equals 5%. Solid lines indicate geometric means. ***P < 0.001 by t-test of log-transformed data. (b) qPCR analysis of vector DNA/mCFTR genomic DNA ratios [geometric means ±95% confidence interval (CI)]. Shown are pUCF2 and pUCF70′ results in FABP mice. The pUL (derivative of pUCF2 with replacement of CO-CFTR with CpG-depleted luciferase) and pUCF42 (derivative of pUCF2 with removed S/MARs and SV40 pA signal inserted between R6K ori and UbC promoter) results in Balb/C mice are shown for comparison. There was no difference at day 14 comparing pUCF2 and pUCF70′, or pair wise among pUCF70′ samples at days 14, 30, and 59 (one-way ANOVA of log-transformed data). (c) Correlation plots between vector DNA/mCFTR genomic DNA (expressed as relative quotients, R.Q.) and the hCFTR/mCFTR mRNA geometric mean ratio at days 14, 30, and 59. There was a significant correlation between vector DNA and CO-CFTR mRNA expression at days 30 and 59 (Supplementary Table S3).

Expression cassette optimization and mRNA and DNA persistence analysis

Numerous attempts including complete CpG depletion of CO-CFTR, utilization of different hCFTR and synthetic introns, use of minicircles or DNA linear fragments, and other strategies failed to maintain CO-CFTR transgene mRNA expression (data not shown). Surprisingly, one pUCF2 derivative, pUCF70 (Figure 2), containing 712 bp of 3′ flanking sequences of the BGH gene in an inverted direction relative to the CO-CFTR coding sequence produced high level and prolonged CO-CFTR expression (236% at day 14 (Figure 3a). This BGH-derived PE region consists of the last 58 codons (176 nucleotides) of the BGH open reading frame, 3′ UTR, BGH polyadenylation signal, and some genomic DNA sequences. Furthermore, the inverted BGH pA orientation suggests that this site is not functional in termination of the CO-CFTR transcript (see Supplementary Figure S4). This study was repeated, designated as pUCF70′. The CO-CFTR expression on day 59 was 14.1%, which was not statistically different than day 30 (52.7%) or day 14 (21.3%) values (one-way ANOVA of log-transformed data). To generate average expression levels for both pUCF70 studies, geometric means from duplicate time points are combined as geometric means (Table 1). Importantly, introduction of the PE element into the pUCF2 plasmid did not reduce CO-CFTR mRNA expression; compare pUCF2 and pUCF70 on day 2 in Figure 3a (t-test of log-transformed data, P = 0.4291).

Figure 2.

Domain analysis of pUCF70. pUCF70 plasmid map and its derivatives, evaluating role of (a) prolonged expression (PE) domain, (b) S/MAR domains, (c) prokaryotic backbone, and (d) promoters.

designates CpG dinucleotides. The PE domain has been removed in pUCF92 and is oriented as a transcriptional terminator in pUCF90. A functional SV40 pA site has been placed 5′ to the PE domain in pUCF106. The interferon-β S/MAR has been removed in pUCF88. The β-globin S/MAR has been removed and replaced by an SV40 pA in pUCF98. Both S/MAR domains have been removed in pUCF100, with placement of an SV40 pA just 5′ to the UbC promoter. pUCF80 is a pUCF70 derivative with the ZeoR-R6K prokaryotic backbone replaced with KmR and the MB1 DNA origin. pUCF108 is a pUCF100 derivative with replacement of ZeoR-R6K by the KmR-MB1 backbone. The UbC promoter has been replaced by the SV40 early promoter in pUCF94, by the human β-actin promoter (and first exon and intron) in pUCF96, and by the CMV promoter and enhancer (and first exon and intron A) in pUCF112.

Table 1. ANCOVA analysis of CO-CFTR mRNA expression levels in the lungs of FABP mice dosed with pUCF70, pUCF80, and pUCF108.

To determine whether the decline in CO-CFTR mRNA expression in pUCF2, or the sustained expression generated by pUCF70, is due to changes in vector DNA persistence, DNA was purified from dosed murine lungs and evaluated by qPCR. The significant decline in measured vector DNA during the first 2 weeks after gene transfer may largely reflect extracellular clearance (Figure 3b). The relative plateau in observed vector DNA from day 14 to day 59 may reflect a progressively larger percentage of nuclear plasmid. This plateau was observed for other compacted plasmids and in both Balb/c and FABP CF mice. If the nuclear plasmid is biologically active, there may be a positive correlation between measured vector DNA and CO-CFTR expression. Correlation analysis was performed on the pUCF70′ time course data (Figure 3c) using the nonparametric Spearman's correlation test (Supplementary Table S3). There was a significant correlation between CO-CFTR mRNA expression and vector DNA on days 30 and 59, but not on day 14. These findings are notable, since a positive correlation between CO-CFTR transgene mRNA and vector DNA in dosed lungs is certainly an anticipated relationship. The finding that this correlation is evident only at long observation intervals suggests that significant time is required to clear extranuclear compacted DNA, which is relatively stable in nuclease enriched environments.

pUCF70 analysis: transcription of the PE region is required for prolonged CO-CFTR expression

Northern blot analysis of pUCF70 transcripts revealed that the PE region, the interferon-β S/MAR, the zeocin resistance gene (ZeoR) prokaryotic backbone, and most of the β-globin S/MAR are transcribed as a 3′ UTR (Supplementary Figure S4). The pUCF70 CO-CFTR transcript terminates in the β-globin S/MAR and is about 8.2 kb, whereas the pUCF2 CO-CFTR transcript is about 5 kb. This finding raised several possible explanations for prolonged CO-CFTR expression by pUCF70. Is prolonged expression related to the presence of the PE region and/or its transcription, the transcription of one or both S/MAR domains, or transcription of the prokaryotic domain with possible influences on heterochromatin formation from otherwise transcriptionally silent plasmid domains in eukaryotic cells? One or more of these factors may be operative in pUCF70. To investigate these possibilities, a series of pUCF70 derivatives (Figure 2) were tested in FABP or FVB/N mice dosed IT with ~100 µg of compacted DNA followed by lung collections at days 2 and 14 (and in some cases at day 30) for quantitative RT-PCR analysis. Geometric means of the day 2 and day 14 hCFTR/mCFTR mRNA ratios, day 14/day 2 percentages, and t-test comparisons for each plasmid are summarized in Table 2.

Table 2. Gene expression comparisons for pUCF70 derivative vectors dosed in FABP (and FVB/N) mice.

We first evaluated whether the PE region and/or its transcription are required. FABP mice were dosed IT with pUCF90 and pUCF92 (Figure 2). In pUCF92, the PE region is removed although this plasmid will still transcribe the full 3′ UTR domains as in pUCF70, thereby permitting a specific evaluation of the necessity of the PE sequence. In pUCF90, the PE sequence is oriented in the opposite direction to recover a functional BGH pA signal, providing a further control for the importance of the PE orientation. As shown in Figure 4, both pUCF90 and pUCF92 generated only transient CO-CFTR expression, with high expression on day 2 (27% and 25%), which fell at day 14 to 0.4% and 1.2%, respectively. These findings indicate that the orientation of the PE region is critical for prolonged CO-CFTR expression in pUCF70.

Figure 4.

Evaluation of pUCF90, pUCF92, pUCF106, pUCF88, pUCF98, and pUCF100 in fatty acid binding protein promoter (FABP) mice. Lungs were dosed intratracheally (IT) with ~100 µg of compacted DNA and harvested at days 2 and 14 for quantitative reverse transcription (qRT)-PCR analysis of CO-CFTR expression. Solid lines indicate geometric means, dotted line equals 5%. **P < 0.01; ***P < 0.001 by t test of log-transformed data.

To evaluate whether active transcription of the PE site is necessary for its ability to generate prolonged CO-CFTR mRNA expression, a plasmid (pUCF106) was prepared that places a functional SV40 pA upstream of the PE site in pUCF70 (Figure 2). pUCF106 was silenced at day 14 (Figure 4), indicating that active transcription of the PE site is required for its ability to generate prolonged expression.

pUCF70 vector modifications: role of S/MAR domains and prokaryotic backbone

In addition to the PE region, active transcription of one or both S/MAR domains might be responsible for improved CO-CFTR mRNA persistence. To evaluate the importance of S/MAR sequences, pUCF88 and pUCF98 (Figure 2), which are pUCF70 derivatives that lack either the interferon-β or β-globin S/MARs, were tested. Both compacted plasmids produced sustained expression (two-way ANOVA of log-transformed data). These findings indicate that either S/MAR appears sufficient for long-term expression, and raises the question whether transcription of any S/MAR domain is necessary.

Two pUCF70 derivatives lacking S/MARs, pUCF100 (containing a CpG-depleted ZeoR-R6K origin backbone) and pUCF108 (containing a CpG-enriched kanamycin resistance gene (KmR)-MB1 origin) (Figure 2) were tested in FABP mice. pUCF100 generated CO-CFTR expression of 51.2% at day 2, which fell to 4.99% at day 14 (P = 0.0007, t-test of log-transformed data) (Figure 4), suggesting that at least one S/MAR domain is required.

However, pUCF108 produced prolonged CO-CFTR expression, with 19.5% on day 2, 9.00% on day 14, and 13.3% on day 30 (Figure 5) (not different by one-way ANOVA of log-transformed data, P = 0.6792). This experiment demonstrates that neither the CpG-depleted ZeoR-R6K backbone nor S/MAR sequences are absolutely necessary for prolonged CO-CFTR expression.

Figure 5.

Time course comparison of CO-CFTR expression by compacted pUCF70, pUCF80, and pUCF180. (a) Evaluation of compacted pUCF108 and pUCF80 in fatty acid binding protein promoter (FABP) mice. Compacted DNA was dosed intratracheally (IT) and animal lungs were harvested at days 2, 14, or 30 for quantitative reverse transcription (qRT)-PCR analysis. Solid lines indicate geometric means, dotted line equals 5%. (b) Comparison of geometric mean ratios from each pUCF70, pUCF80, and pUCF108 FABP mouse study. Linear regression analysis of log-transformed data was used to generate best-fit lines. A statistical comparison (Table 1) of slope and y-intercept values using an analysis of covariance (ANCOVA) generated a nonsignificant difference in slope (P = 0.6344) and y-intercept (P = 0.08445), although the nearly 5–10-fold improvement in geometric mean ratios for pUCF70 compared to pUCF80 and pUCF108 at days 2, 14, and 30 is noteworthy. These findings suggest that pUCF70 may be superior to pUCF80 and pUCF108 if the sample size of the analysis was increased. Dotted line equals 5%. Error bars = SEM, n = 6–13 per time point.

Another pUCF70 derivative, pUCF80 (Figure 2), containing the KmR-MB1 backbone insulated with two S/MARs, was tested in two-independent studies (Figure 5). Combined pUCF80 results are presented in Table 1, and data analysis did not reveal silencing. However, CO-CFTR expression levels produced by both plasmids containing the KmR-MB1 backbone (pUCF80 and pUCF108) were lower at each time point compared with pUCF70 (Figure 5b and Table 1), which nearly reached statistical significance by analysis of covariance (ANCOVA) analysis.

pUCF70 vector modifications: promoter selection influences expression duration profile

To determine whether the UbC promoter is necessary for pUCF70 to produce prolonged CO-CFTR mRNA expression, pUCF70 derivative plasmids were prepared replacing UbC with either the human β-actin (pUCF96), SV40 early (pUCF94), or CMV immediate early (pUCF112) promoters (Figure 2). Additionally, pUCF96 and pUCF112 have replaced the UbC first intron with the human β-actin first intron or CMV intron A, respectively; the UbC first intron remains intact in pUCF94. Both pUCF96 and pUCF94 generated sustained CO-CFTR expression, whereas pUCF112 was silenced (Figure 6a,b). These data indicate that the UbC promoter can be replaced by other promoters in the pUCF70 vector and still achieve maintained expression, although some promoters, such as CMV, will be silenced.

Figure 6.

Evaluation of alternative promoters in the pUCF70 vector. (a) The UbC promoter and first exon/intron were replaced with the human β-actin promoter and first exon/intron (pUCF96) or the UbC promoter was replaced with the SV40 early promoter (pUCF94). Plasmids pUCF96 (108 µg) and pUCF94 (104 µg) were dosed intratracheally (IT) in fatty acid binding protein promoter (FABP) mice and harvested at days 2 and 14 for quantitative reverse transcription (qRT)-PCR analysis. For comparison are prior data with pUCF70 and the day 14 repeat pUCF70′ study. There was no statistically significant difference between pUCF94 and pUCF96 (two-way ANOVA of log-transformed data). (b) The prolonged expression (PE) domain does not facilitate prolonged CO-CFTR mRNA expression transcriptionally controlled by the CMV promoter/enhancer (and also containing the CMV first exon and intron A). Compacted pUCF112 was dosed IT (113 µg) in FVB/N mice and harvested on days 2 and 14 for qRT-PCR analysis. For comparison are pUCF70 results in FVB/N mice. Solid lines indicate geometric means, dotted line equals 5%. ***P < 0.001 by t test of log-transformed data.

Discussion

The focus of this study is to optimize an hCFTR expression vector for compacted DNA nanoparticle therapy of the lungs of CF patients. Our objectives are to produce hCFTR transgene mRNA in the lungs of mice at levels >5–10% of endogenous mCFTR and to have hCFTR expression persist for multiple weeks after a single dose. To potentially improve both mRNA and protein expression, we developed a hCFTR synthetic gene that was codon-optimized, CpG-depleted (except for one 3′ CpG), has removed 5′ and 3′ UTRs of natural hCFTR, and has an improved Kozak consensus sequence. When tested in HEK293 cells for protein expression, CO-CFTR (pUCF2) produced ninefold more hCFTR protein than hCFTR cDNA (pUCF) (Figure 1b). When dosed in murine lungs as DNA nanoparticles, pUCF2 generated higher expression than pUCF in Balb/c (14.6-fold) and FABP (3.7-fold) mice (Figure 1c,d). Although we do not have specific evidence for the importance of any given CO-CFTR modification, these mRNA and protein findings strongly suggest that CO-CFTR is transcribed and/or processed and translated more efficiently than hCFTR cDNA in the context of these vectors.

Two days after transfection of FABP CF mice lungs, expression of CO-CFTR mRNA ranging from 10 to 100% were observed for pUCF2 and multiple vector derivatives. However, in each of these vectors day 14 expression fell significantly, below the 5% benchmark. We show that this decline in CO-CFTR expression is not due to vector DNA loss or immunogenicity of hCFTR, and thus is likely due to transcriptional silencing. Multiple strategies were employed to potentially address vector silencing mechanisms, including an evaluation of promoters, minicircles, linear fragments, and use of chromatin barrier elements flanking the eukaryotic cassette. None of these strategies proved to be successful, with all of these compacted DNAs producing <5% CO-CFTR/mCFTR RNA geometric mean ratios by day 14. When evaluating various pA sites, a BGH gene 3′ flanking sequence containing the BGH pA signal in a reverse and presumed inactive orientation had an unexpected result, and was named the PE domain. CO-CFTR mRNA expression was sustained in the lungs of multiple murine strains dosed with compacted pUCF70, which contains this PE domain. A time course study in FABP mice (Figures 3a and 5b) produced sustained CO-CFTR mRNA for 59 days, meeting our preclinical expression objectives. Importantly, the 53% hCFTR/mCFTR mRNA ratio at day 30 suggests the potential for significant hCFTR biological activity for at least 1 month after dosing.

An analysis of the pUCF70 CO-CFTR transcript revealed an extended 3′ UTR including the PE sequence, the interferon-β SMAR, the ZeoR prokaryotic backbone, and a portion of the β-globin S/MAR (Supplementary Figure S4). The transcript terminated in the β-globin S/MAR, which contains several transcriptional terminators. This result suggested several mechanisms that might account for sustained CO-CFTR expression, including transcription of the PE domain, transcription of one or both S/MARs, transcription of the prokaryotic backbone, and complex interactions requiring one or more of these processes.

We demonstrated that the presence of the PE sequence in a specific orientation and its transcription are necessary for prolonged CO-CFTR mRNA expression in pUCF70. pUCF70 derivatives without the PE domain (pUCF92) or with the PE domain in a reversed orientation (pUCF90) were silenced by day 14 (Figure 4). Placement of a functional SV40 pA site just upstream of the PE site (pUCF106) also produced silencing at day 14, thus demonstrating the necessity of transcription of the PE region for persistent CO-CFTR expression.

In contrast to the PE sequence, transcription of S/MAR domains is not always required for prolonged CO-CFTR expression. This result was somewhat surprising, since S/MAR domains have multiple properties, including attachment to the nuclear matrix at sites that appear favorable for mRNA transcription and processing.²⁰ However, the requirement for S/MAR domains is influenced by the specific prokaryotic backbone, as illustrated by plasmids pUCF100 (ZeoR-R6K) and pUCF108 (KmR-MB1). S/MARs may be needed when the ZeoR-R6K backbone is employed. A detailed time course analysis of pUCF70, pUCF108, and pUCF80 (Figure 5b and Table 1) illustrates these features and also demonstrates that substitution of the KmR-MB1 backbone for ZeoR-R6K reduces levels of CO-CFTR expression to near statistical significance. Therefore, there appears to be a complex dynamic between various plasmid domains and function of the PE sequence to generate sustained CO-CFTR expression. This dynamic emphasizes the importance of specific empiric tests of a candidate expression plasmid design, and that substitution of seemingly unrelated components, such as the prokaryotic backbone, can have unexpected and significant affects on transgene levels and expression duration.

We also considered that transcription of the prokaryotic backbone as a 3′UTR in the CO-CFTR transcript may be important in diminishing probabilities of heterochromatin formation in pUCF70 DNA. Improved duration of expression by minicircles compared to plasmids in the liver may relate to removal of transcriptionally silent DNA domains that are prone to nucleosome formation, with subsequent heterochromatin encroachment into the eukaryotic transcriptional cassette resulting in transgene silencing.²¹ It is noteworthy, however, that evaluation of CO-CFTR minicircles and linear fragments in the mouse lung were not successful in generating prolonged CO-CFTR expression (data not shown). Moreover, active transcription of the ZeoR prokaryotic backbone in plasmid pUCF100, which contains the PE region but no S/MARs, was unsuccessful in generating sustained CO-CFTR expression (Figure 4). This finding suggests that active transcription of the prokaryotic backbone in the mouse lung may not have a significant role in facilitating prolonged CO-CFTR expression.

Importantly, we show that prolonged CO-CFTR expression in pUCF70 is not dependent on the UbC promoter. This feature provides opportunities for cell type-specific expression in the lung and other target organs, and expands the potential of the PE element to optimize expression duration for other transgenes that are difficult to express long-term.

In summary, we identified a new genetic element, the PE sequence, that allows sustained CO-CFTR expression using many different vectors. This element is successful in facilitating prolonged CO-CFTR mRNA expression in plasmids containing either the CpG-depleted ZeoR-R6K or the CpG-enriched KmR-MB1 backbone, and promoters other than UbC can be employed. Chromatin barrier S/MAR domains and CpG-depletion are not required. The mechanism by which the PE site facilitates prolonged CO-CFTR mRNA expression remains undefined and will require further investigation. Our studies indicate that the PE sequence must be in a specific orientation and transcribed in order to facilitate sustained expression; it appears nonfunctional when present in the plasmid as a nontranscribed element. These findings highlight the possibility that the PE region may bind other transcriptional control factors as an mRNA structural element. Deletion analysis may provide more insight about mechanisms underlying the PE sequence. Regardless of the mechanism(s) involved, we have shown that plasmids can be developed which generate clinically relevant levels of hCFTR mRNA expression for prolonged duration in the mouse lung.

Materials and Methods

DNA vector construction. DNA vectors were constructed using standard molecular biology techniques.²² Each DNA fragment that was amplified by PCR or synthesized or obtained from a third party was sequenced. All DNA junction regions were sequenced. Plasmid integrity was confirmed by restriction analysis using at least four restriction endonucleases. Unless otherwise specified, molecular biology reagents were either from Applied Biosystems (Carlsbad, CA) or Ambion (Austin, TX). For details of plasmid construction, see Supplementary Materials and Methods. The CO-CFTR and PE domain sequences are referenced in Genbank as GN114138.1 and GN114137.1, respectively.

Compacted DNA preparation. Compacted DNA was manufactured by adding the DNA solution at a controlled rate to a vortexing tube of PEGylated polylysine (CK₃₀PEG10k, acetate counterion).^6,11 CK₃₀PEG10k and DNA were formulated at a final charge ratio of 2:1. The DNA working stock (20 ml, 0.1 mg/ml) was added to 2.0 ml of CK₃₀PEG10k (3.2 mg/ml) stock at a rate of 4 ml/min by a syringe pump. During this addition, the tube of CK₃₀PEG10k was vortexed at a controlled rate so that the two materials mixed instantaneously. The compacted DNA was then filtered through a vacuum driven sterile filter with 0.2 µm polyethersulfone membrane (Millipore, Billerica, MA).

The filtered sample of compacted DNA was then concentrated 20–30-fold using Vivaspin centrifugal concentrators (MWCO 100k) (Sartorius Stedim, Bohemia, NY). The concentrated DNA was then diluted 20–30 fold with 0.9% NaCl to remove excess CK₃₀PEG10k and exchange solvents with physiologic saline. Then the compacted DNA was concentrated again 20–30-fold, to a final concentration of 2–4 mg/ml. After formulation, the compacted DNA underwent several quality control tests, including sedimentation, turbidity, gel electrophoresis, and transmission electron microscopy. These methods are described in refs. 6,11. Also, endotoxin levels were measured using an Endosafe Portable Test System manufactured by Charles River Laboratories (Wilmington, MA).

Intranasal and IT dosing in mice. Each intranasally dosed mouse was given an intraperitoneal injection of 100–150 µl anesthetic cocktail (8.6 mg/ml ketamine, 0.29 mg/ml acepromazine, and 1.7 mg/ml xylazine). An EDP-Plus electronic pipette (Rainin Instrument, Oakland, CA) was programmed to deliver 10 × 2.5 µl aliquots, for a total of 25 µl of compacted DNA. The anesthetized mouse was hand-held, ventral side up. The bottom lip of the mouse was gently pushed up over the mouth, sealing the mouth shut. The electronic pipette was quickly placed on the bridge of the nose and 10 doses of 2.5 µl compacted DNA were given, for a total administered dose of 25 µl. The compacted DNA was at a concentration of 4 mg/ml in saline, for a total dose of ~100 µg per mouse. The doses were given at such a rate that the mouse inhaled the droplets before large droplets were allowed to accumulate on the nose. After administration, the mouse was immediately placed ventral-side down and allowed to recover. A heating blanket was used for 24 hours under the cage to aid in recovery and to prevent anesthesia-related hypothermia.

For IT dosing, each mouse was given an intraperitoneal injection of 100–150 µl of the anesthetic cocktail listed above. The mouse hair on the front of the neck was removed and betadine solution was applied to the area. Each mouse was situated on a surgical board at a 30% incline¹¹ and a tracheostomy was performed with insertion of a 22-gauge catheter (Becton Dickinson, Franklin Lakes, NJ). A 50-µl aliquot of compacted DNA was administered as a bolus into the trachea using a manual pipette. For IT dosing, the compacted DNA was at a concentration of 2 mg/ml in saline, for a total dose of ~100 µg/mouse. After the bolus was delivered, the wound was covered with betadine solution. The mouse was immediately placed ventral-side down and allowed to recover. A heating blanket was used for 24 hours under the cage to aid in recovery and to prevent anesthesia-related hypothermia.

All procedures were conducted in strict compliance with protocols as approved by the Institutional Animal Care and Use Committee at Case Western Reserve University School of Medicine.

Preparation of total RNA and cDNA synthesis. Mouse lungs harvested for reverse transcriptase (RT)-PCR were collected in 2 ml tubes containing a 5-mm stainless steel bead (Qiagen, Valencia, CA) and 1 ml of 1× Nucleic Acid Lysis Solution (Applied Biosystems). Samples were immediately placed on dry ice and kept at -80 °C. Frozen lungs were thawed on wet ice for 30 minutes before homogenization. Homogenization was performed using a TissueLyser (Qiagen) for 4 minutes at 30 Hz, a total of three times, cooling on ice for 5 minutes between homogenizations. Dilution of the homogenate was made by adding 50 µl of homogenate to 750 µl of 1× Nucleic Acid Lysis Solution (Applied Biosystems). Total RNA was isolated from the lungs using a 6100 Nucleic Acid PrepStation (Applied Biosystems) as recommended by the supplier. Three hundred microliter of the diluted and filtered homogenate was used for the isolation. Total RNA was treated with Turbo DNA-free DNase (Ambion) to remove contaminating DNA (37 °C for 40 minutes).

The reverse transcription reaction was performed using the High-Capacity cDNA RT Kit (Applied Biosystems) as recommended by the supplier. Minus RT reactions were set up, substituting nuclease-free water for Multiscribe Reverse Transcriptase (Invitrogen, Carlsbad, CA) in each reaction. All RT reactions were diluted 1:1 with nuclease-free water.

Real-time quantitative TaqMan RT-PCR. Quantitative RT-PCR assays were performed using the ABI Prism 7300 Real-Time PCR System and Sequence Detection V1.3.1 software (Applied Biosystems). Fluorogenic probe and oligonucleotide primer combinations for TaqMan assays were designed using Primer Express V3.0 (Applied Biosystems). The TaqMan CO-CFTR RT-PCR primer set K6, forward (5′TGTGCTGAGCAAGGCCAAG′), reverse (5′CAGGGTTCTTCTGATGATCTGGTAG) and Fam/Tamra probe K (5′-FAM CTCTGCCCACCTGGACCCTGTGA) were specific to CO-CFTR mRNA. The TaqMan mCFTR RT-PCR assay primer set B8 forward primer (5′CTAGTCCATTCCCAGAACCCAT), reverse (5′GGGATCCACCTGTCTCTGTGTC), and Fam/Tamra probe B (5′-FAM AGGCATTTCCCATGCTTCTAACCCCA) detected endogenous mouse CFTR mRNA. The TaqMan mGAPDH RT-PCR primer set A forward primer (5′TGGCCTCCAAGGAGTAAGAAAC), reverse primer (5′GGGATAGGGCCTCTCTTGCT), and Fam/Tamra probe A (5′-FAM ACCACCCACCCCAGCAAGGACAC) detected mouse GAPDH, which was used as a normalizer. The relative quantitation mode was used in all RT-PCR assays, which contained 900 nmol/l of forward primer, 900 nmol/l of reverse primer, and 250 nmol/l probe in a total volume of 25 µl, of which 5 µl was cDNA. All samples were assayed in quadruplicate, including the minus RT reactions. Rarely contamination was detected, but ignored because the difference between plus and minus RT signals was more than 6 Cts.

Immunoprecipitation/western analysis. HEK293 cells were plated onto 60 mm dishes and transfected with either 0.75 µg or 3 µg of plasmid using Lipofectamine reagent (Invitrogen). To minimize differences in transfection efficiency between plates, each liposome/DNA transfection mixture contained an equal total amount of plasmid DNA, equal pmols of each test plasmid (but differing microgram amounts), an equal amount of luciferase plasmid (10 ng) to assess transfection efficiency, and appropriate amounts of “filler ” plasmid (Bluescript) (Agilent Technologies, Santa Clara, CA). Cells were collected on day 2 post-transfection and lysed in Buffer A (50 mmol/l Tris, pH 7.4, 150 mmol/l NaCl, 1 mmol/l EDTA) including 1% Triton X 100 and protease inhibitors (Roche Complete Mini Protease Inhibitor Tablets; Roche Diagnostics, Indianapolis, IN) for 30 minutes on ice. The lysate was centrifuged at 16,000g for 5 minutes at 4 °C and the supernatant was evaluated for luciferase activity and stored at -80 °C until further analysis. Rec-Protein G sepharose 4B (Invitrogen) was incubated with the immunoprecipitating anti-CFTR antibody (#MAB1660; R&D Systems, Minneapolis, MN) overnight at 4 °C while rocking. The resin-antibody complexes were washed with Buffer A including 0.1% Triton X 100, and incubated with the transfected cell culture supernatant overnight at 4 °C while rocking. The complexes were collected by pulse centrifugation and washed with Buffer A including 0.1% Triton X 100 and protease inhibitors. The complexes were resuspended in 2× sample loading buffer and incubated at 95 °C for 5 minutes. The sample supernatants were run on a 4–20% SDS Tris–glycine gel (Invitrogen, Carlsbad, CA) and transferred to Protran nitrocellulose paper (Whatman, Piscataway, NJ) using the Multiphor II Transfer System (Pfizer, New York, NY). The nitrocellulose blot was washed briefly in TBS (Sigma, St Louis, MO) and blocked in SuperBlock T-20 (Thermo Scientific, Rockford, IL). The blot was incubated with the mouse monoclonal anti-CFTR antibody (#MAB1660; R&D Systems, Minneapolis, MN) followed by ECL anti-mouse IgG horseradish peroxidase (#NA391V; GE Healthcare/Amersham, Piscataway, NJ). Horseradish peroxidase detection was through the SuperSignal West Femto Maximum Sensitivity Substrate Kit (Thermo Scientific) and the blot was exposed to the Fluor S Multi Imager (Bio-Rad, Hercules, CA) for 5 minutes. Band densities were quantified and normalized for modest differences in luciferase activity.

Statistical analysis. For analysis and for presentation in figures and tables, mRNA data (hCFTR or CO-CFTR/mCFTR) and DNA data (vector DNA/mCFTR genomic DNA) were log-transformed in order to equalize variances and make distribution of data points more Gaussian. Consequently, comparisons are made between geometric means of the data and 95% confidence interval and SEM are shown on a logarithmic scale. Ratios of mCFTR mRNA to mGAPDH mRNA in various mouse strains were not log-transformed since in all cases but one the data conformed to normal distribution (by D'Agostino and Pearson omnibus normality test). Data for gene expression on day 2 and day 14 after transfection were analyzed by the two-tailed, unpaired t-test. In cases where unequal variances were found (via F test), Welch's correction was applied. When data for one plasmid and more than two time points were compared, a one-way ANOVA with Bonferroni's post-test was used. When data for more than one plasmid and two time points were analyzed simultaneously, a two-way ANOVA was used. Covariation of CO-CFTR expression and amount of vector DNA in lung tissue was evaluated by calculating the nonparametric Spearman correlation. Comparison of expression level and its longevity for plasmids pUCF70, pUCF80, and pUCF108 was done by first performing linear regression analysis of log-transformed geometric means for each plasmid at different time points and then comparing slopes and y-intercepts of generated best-fit lines using an analysis of covariance. For all these tests the threshold for statistical significance was set at P = 0.05.

Supplementary Material

Figure S1.

Standard curves for CO-CFTR and mCFTR were generated using a validation template vector containing both genes.

Figure S2.

Endogenous lung mCFTR/mGAPDH expression ratios.

Figure S3.

Compacted pUCF2 was dosed IT (~100 µg) into FABP, 129, C57, FVB/N, and SCID mice.

Figure S4.

Northern blot of HEK293 cells transfected with pUCF70, pUCF2, and pUL.

Table S1.

Adjustment of CO-CFTR Ct values based on standard amplification curve analysis using vectors containing CO-CFTR and mCFTR templates.

Table S2.

Expression of endogenous mCFTR in different mouse strains.

Table S3.

Correlation analysis of mRNA and vector DNA levels in the FABP lung: pUCF70 time course (study pUCF70′).

Materials and Methods

.